Research on the Recognition Algorithm for the Presence of Occlusion in Face Images based On DDPM

In recent years, with the development of AI, face recognition technology has made great progress. However, there are still many challenges and problems with face recognition in complex environments. Especially in the era after the new crown, new viruses appear from time to time, and wearing masks has become a daily need for people. Masks obscure the key features of the face, and traditional face recognition technology cannot correctly recognize faces in access control systems, time and attendance systems, and security systems. Therefore, it is very meaningful to study how to recover the masked face image into a real image without mask, which is helpful to improve the accuracy of face recognition. In this article, for the research of local occlusion face recognition technology, we propose the face image restoration technology based on the diffusion model to remove glasses and masks, which solves the problem of effectively restoring the details of the face under the large-area occlusion. Finally, the restored face is input into Resnet network to achieve the purpose of face recognition. The experiment proves that the use of diffusion model denoising face can effectively improve the accuracy of recognition.